Using small-magnitude earthquakes to investigate the interplay between seismic and aseismic deformation along the Hellenic Subduction System

The Hellenic Subduction System (HSS) in the eastern Mediterranean is the oldest active subduction margin on earth. It is a segmented boundary that hosts the continuum of faulting styles over a ~200km range in depth and can generate large earthquakes with high tsunamigenic potential.  The complexity of deformation styles and rates leave key aspects of the system poorly understood. For example, historical records of Mw<8 earthquakes fail to explain the current observed convergence rate (~35mm/year), and recent geodetic measurements suggest that the degree of locking within the system is heterogeneous. The density of geodetic measurements is increasing rapidly, nevertheless, the inherent time lag required to accumulate data that will enable identifying regions that undergo slower (than seismic) deformation transients will necessitate inferences from seismic signals. In this work, we aim to further close the observational gap between heterogeneous deformation styles and rates using the features of seismicity distributions to infer where deformation rates, and by inference, locking, vary most.    To that scope, we will present new results of an enhanced earthquake catalog that we will use to explore the spatio-temporal distribution of seismicity features (e.g., b-value, effective stress drop, seismic-moment-release skewness) to infer variability in deformation rates and loading. Catalog enhancement exploits data from the temporary (EGELADOS) broadband seismometer network that operated between 2005 until 2007 combined with permanent stations leading to a station spacing of ~40 km and covering the entire southern Aegean Sea. We first use the combined network to detect earthquakes using machine learning approaches (EQTransformer, PhaseLink) for detection, phase picking and association. After performing initial locations using NonLinLoc combined with a 1D velocity model and quality control procedure, we enhance the number of small-magnitude detections using a multi-station template-matching approach. Next, we scan the enhanced high-resolution catalog for distinct spatial and temporal patterns of seismicity using unsupervised clustering. We then quantify the clustered seismicity using b-value, effective stress drop, and seismic-moment-release skewness (among other parameters). We will present our clustering results in the context of the variability in slip phenomena related to earthquake-earthquake interactions (e.g., static and dynamic triggering) as well as in the context of external forcing (e.g., aseismic triggering or fluid migration).   The preliminary results that we will present will provide a basis for our more broad-scale study of interplay between seismic and aseismic deformation. In particular, where the latter is gradually becoming increasingly resolvable using GNSS data within the HSS, this work will provide a basis for links with geodetically observed deformation in the future.